Publication number  US7231007 B2 
Publication type  Grant 
Application number  US 10/450,610 
PCT number  PCT/FI2001/001134 
Publication date  Jun 12, 2007 
Filing date  Dec 19, 2001 
Priority date  Dec 22, 2000 
Fee status  Paid 
Also published as  US20040042532, WO2002052750A1 
Publication number  10450610, 450610, PCT/2001/1134, PCT/FI/1/001134, PCT/FI/1/01134, PCT/FI/2001/001134, PCT/FI/2001/01134, PCT/FI1/001134, PCT/FI1/01134, PCT/FI1001134, PCT/FI101134, PCT/FI2001/001134, PCT/FI2001/01134, PCT/FI2001001134, PCT/FI200101134, US 7231007 B2, US 7231007B2, USB27231007, US7231007 B2, US7231007B2 
Inventors  Atte Artamo, David Astély 
Original Assignee  Nokia Corporation 
Export Citation  BiBTeX, EndNote, RefMan 
Patent Citations (7), NonPatent Citations (1), Referenced by (14), Classifications (7), Legal Events (4)  
External Links: USPTO, USPTO Assignment, Espacenet  
The invention relates to a measuring method employing a spatial diversity antenna and to a receiver.
Interference caused by other users or systems and noise sum into a signal on the radio path. Radio systems therefore employ different diversity methods to increase the coverage area and/or capacity of the system. One of them is spatial diversity, which is obtained using an array antenna comprising a plural number of antenna elements that are physically separate from each other. The received signals are combined in diversity receivers that employ maximalratio combining, for example.
Current receivers are based on statistical signal models the accuracy of which cannot be relied on in all situations. The impact of noise and interference can be reduced in the Maximal Ratio Combining (MRC) method, for example. However, this method supposes that the interference and noise in each antenna element are independent of other antenna elements, i.e. they are white. In actual cellular radio networks this is not always true; for example, the interference that highpower signals, a few at the most, cause upon reception may affect all the antenna elements, i.e. the interference in the antenna elements is coloured.
Interference Rejection Combining (IRC) is a method that does not contain assumptions about whether interference and noise correlate with antenna elements. A disadvantage of the IRC method is that it occupies more baseband processing capacity than the MRC method, because there are more parameters to be estimated although the sample group available is the same. When used extensively, baseband processing may reduce the ability of the base station to serve the subscriber terminals and, thereby, it may impair cell capacity. Due to errors that may take place in parameter estimation if the number of parameters to be estimated is large, the use of complex interference elimination methods may even degrade the capacity of the receiver.
It is therefore an object of the invention to provide an improved measuring method and an arrangement implementing the method to avoid the use of complex interference elimination methods when they are not needed. In addition, the aim is to enhance radio system design and maintenance. These objects are achieved by a measuring method for a receiver in a radio system, the receiver comprising an array antenna for signal reception and the array antenna comprising at least two antenna elements. The method comprises the steps of measuring with the array antenna the spatial colour of received interference; controlling a method to be used for combining the signals received by the different antenna elements on the basis of the measurement result of the spatial colour of the interference.
The invention further relates to a measuring method for a receiver in a radio system, the receiver comprising an array antenna for signal reception and the array antenna comprising at least two antenna elements. The method comprises the steps of measuring with the array antenna the spatial colour of received interference; monitoring the spatial colour of the interference and taking into account the result of the colour measurement for enhancing the operation of the radio system.
The invention further relates to a receiver comprising an array antenna for signal reception, the array antenna comprising at least two antenna elements. The receiver further comprises an interference detector for measuring the spatial colour of the interference received with the array antenna; and controlling means for controlling a method to be used for combining the signals received by the different antenna elements on the basis of the measurement result of the spatial colour of the interference.
The invention still further relates to a receiver comprising an array antenna for signal reception, the array antenna comprising at least two antenna elements. The receiver further comprises an interference detector for measuring the spatial colour of the interference received with the array antenna; and monitoring means for monitoring the spatial colour and for using the measurement result of the spatial colour of the interference for enhancing the operation of a radio system.
The preferred embodiments of the invention are disclosed in the dependent claims.
The invention is based on deciding on a signal model suitable for a received signal on the basis of a sample group taken from the signal. In a hard decision mode it is decided whether a spatially coloured or spatially white model of noise and interference will be created. In a soft decision mode the degree of spatial colour in the interference is determined.
The method and arrangement of the invention provide several advantages. With the statistical model of the received signal known, it is easier to estimate a covariance matrix for noise and interference. The statistical signal model allows assumptions to be made regarding the existence of strong interference sources.
In the following, the invention will be described in greater detail with reference to the preferred embodiments and the accompanying drawings, in which
The described solution may be applied in digital TDMA, FDMA and CDMA radio systems, although it is not restricted to them.
Let us first examine the theoretical basis of the disclosed solution. A received signal can be expressed as
r(t)=h _{k,m} _{ s } _{k}(t−τ _{m,k})+u _{k,m}(t) (1)
where r(t) is a vector corresponding to the received signal and comprising L elements, L denoting the number of diversity antenna elements, M is the number of paths, m is a path index, s(t) is a transmitted symbol, τ_{m }denotes a propagation delay and h_{m }denotes a complex channel response on the m^{th }path, and vector u(t)=(u_{1}(t),u_{2}(t), . . . ,u_{L}(t))^{T }represents disturbance that consists of interference and noise. Interference mainly consists of multiple access interference (MAI) and interpath interference (IPI), for example. Noise, in turn, may be thermal noise associated with the antenna structure, for example. Interference can be expressed with the following vector:
where K is the number of users, M_{k }the number of paths of k^{th }user, h_{k,m }is the channel estimate for the m^{th }path of the k^{th }user, n(t) is noise and S_{k }is the symbol of the k^{th }user, sent on the m^{th }channel. User k=0 is thought to represent the desired signal, the signals of the rest of the users being thought of as interference. Noise n(t) represents uncorrelated interference appearing in the antennas. A spatial covariance matrix R_{rr }of the received signal r(t) may be expressed as follows:
R _{rr} =E
However, since spreading coding is not used in the TDMA and FDMA, this approximation cannot be made in them, but the covariance matrix R_{uu }of formula (4) must be formed of samples from which the desired signal has been removed.
Let us now examine the estimation of the covariance matrix R_{rr }in greater detail with particular reference to the CDMA system. The spatial covariance matrix R_{uu }for interference (R_{uu}≈R_{rr }is valid in the CDMA system) indicates the degree of similarity of the interference in the different antenna elements, i.e. the correlation of interference between the different antenna elements. If interference signals received by the different antenna elements do not correlate with each other, the interference is spatially white and its covariance matrix R_{uu }can be expressed as follows:
R _{uu}=diag(σ_{1} ^{2},σ_{2} ^{2}, . . . ,σ_{L} ^{2}), (7)
where σ_{1} ^{2 }denotes the intensity of the interference in an I^{th }antenna element. If, in addition, the intensity of the interference received by each antenna is the same, the following expression is possible:
R _{uu}=σ^{2} I, (8)
where I is a unit matrix.
On the other hand, if the interference signal received by the different antenna elements is coloured, i.e. there is correlation between the interference signals of the different antenna elements, the covariance matrix can be written as follows:
where σ_{1} ^{2 }denotes the product of interference u_{1}u_{i} ^{* }in a first antenna element, u_{i} ^{* }representing a complex conjugate of u_{i}. In a broad sense, the received signal r(k) is modelled as a stationary process during the sampling and therefore its covariance matrix R_{rr }represents a deterministic parameter. Table 1 shows a summary of different covariance matrixes R_{rr}, which in the CDMA system corresponds basically to R_{uu}. On the basis of a sample of a received signal it is possible to conclude which covariance matrix model of those shown in Table 1 is the most applicable for modelling the signal.
TABLE 1  
Models for covariance matrix R_{rr}  
1. The received signal isspatially white, its powerbeing the same in all theantenna elements 


2. The received signal isspatially white, its powerbeing different in differentantenna elements 


3. The received signal isspatially white 


The solution disclosed herein will concentrate on models 1 and 3. Under the impact of spatially coloured interference, the covariance matrix of the interference is nondiagonal. This means that the interference detected at reception correlates in the separate receiving antenna elements during the time the covariance matrix R_{uu }is formed. Uneven distribution of interference in receiving directions may be due to a number of reasons, such as a high data transfer rate or power of a signal received from a subscriber terminal compared with other transmitters, a malfunction of a subscriber terminal, an illegal transmitter, a signal from a transmitter of another radio system (because of poor network planning), etc.
The received signal is supposed to follow normal distribution and the samples are temporally white, i.e. samples taken at different points of time are mutually independent. Consequently, the following probability density function (PDF) common to P samples of the received signal r[p], p=1 . . . P, is arrived at:
From this, the following loglikelihood function is obtained:
−ln f({r[p], p=1 . . . P}, Θ)=PL lnπ+PlnR _{rr}(Θ)+Ptrace{{circumflex over (R)} _{rr} R _{rr } ^{−1}(Θ)} (11)
where trace{{circumflex over (R)}_{rr}R_{rr} ^{−1}(Θ)} represents the trace of matrix {circumflex over (R)}_{rr}R_{rr} ^{−1}(Θ), i.e., the sum of diagonal elements, matrix R_{rr }denotes an expected value obtained from the received signal, a covariance matrix {circumflex over (R)}_{rr }estimated on the basis of the received samples being defined as follows:
where vector r[p] representing the signal is the p^{th }sample of the received signal. In the disclosed solution, the number K of samples must be greater than the number L of antennas, i.e. K≧L. In this case the covariance matrix has an inverse matrix, which is necessary in order for a logarithmic, concentrated likelihood function to be formulated.
Let us then examine the deducing of an ML estimate for covariance matrix R_{rr}, assuming that the received signal is spatially white. Interference signals between the different antenna elements are in this case assumed to be uncorrelated, and the negative loglikelihood function is:
By setting the value of the loglikelihood function derivative to zero, in order to obtain a local minimum, i.e.
the following ML (Maximum Likelihood) estimate is obtained for the interference signal variance σ^{2}:
The concentrated cost function obtained is
Logarithm In used in the function is a natural logarithm, its base number being e (e≈2.71828182 . . . ).
Let us then examine the deducing of the ML estimate for covariance matrix R_{rr}, assuming that the received signal is spatially coloured. In this case the interference signals between the different antenna elements are assumed to be correlated. The negative loglikelihood function now takes the following form:
By setting the value of the loglikelihood function derivative to zero, in order to obtain a local minimum, i.e.
the following ML estimate is obtained for the covariance matrix of the samples:
R={circumflex over (R)}_{rr} (19)
Hence, the concentrated cost function obtained is
L({circumflex over (R)})=PL lnπ+P ln{circumflex over (R)} _{rr} +PL (20)
Table 2 provides a summary of the models for the MRC and IRC combining methods, ML estimates and the concentrated ML cost functions for different signal models.
TABLE 2  
Models, ML estimate and ML cost functions  
Concentrated ML cost  
function  
Model  ML estimate  −ln f({r[p], p = 1 . . . P}, {circumflex over (Θ)}) 






The concentrated cost function is needed for deducing the method of detecting the spatial colour of interference.
The spatial colour of interference is detected using an interference detector, which can be used for controlling the receiver to select a combining method. Examples of the combining methods include MRC combining and IRC combining. These combining methods correspond to different ways of weighting the signals arriving from the different antenna branches, before they are combined. The interference detector can also be used for monitoring the spatial colour of interference and for using the monitoring results in connection with the maintenance or design of a radio system to provide enhanced radio system operation.
A basic principle is to select a combining method based on the most appropriate covariance matrix model in each case. The most appropriate covariance matrix model is selected using the MDL (Minimum Description Length) criterion. Covariance models, which comprise varying numbers of parameters, are compared with each other on an equal basis by applying the MDL criterion, the best covariance model being the one that minimizes the following expression:
MDL(M)=−2 ln f({r[p], p=1 . . . P}, {circumflex over (Θ)})+M lnP (21)
where M is the number of parameters, either 1 or L^{2}, in parameter variable Θ. An increase in the number of parameters results to an estimate corresponding increasingly better to reality and to a decrease in the negative loglikelihood function. The second term represents a penalty term that grows as the number of parameters increases.
Table 3 shows the MDL cost function for the two models (MRC and IRC).
TABLE 3  
MDL cost functions  
Number of  
parameters  MDL  
M  −2ln f({r[p], p = 1 . . . P}, {circumflex over (Θ)}) +Mlnp  
MRC  1 


IRC  L^{2} 


In the hard decision mode, when either IRC or MRC is to be used, the following selection rule is obtained for the combining method:
In other words, when the MDL cost function corresponding to the IRC method produces a lower value than the MDL cost function corresponding to the MRC method, the IRC method is selected in the hard decision mode. Correspondingly, the MRC method is selected when the MDL cost function corresponding to the MRC method produces a lower value than the MDL cost function corresponding to the IRC method.
The selection can be simplified into
The IRC method is thus selected when the MDL cost function
produces a lower value than the MDL cost function corresponding to the MRC method. Correspondingly, the MRC method is selected when the MDL cost function
corresponding to the MRC method produces a value that is lower than (or at most equal to) the value provided by the MDL cost function f_{IRC}=
corresponding to the IRC method. The computation can be simplified because the righthand side term
can be computed in advance for the desired solution, the righthand side term being only dependent on the number L of the antennae and the number P of the samples. In the soft decision mode, the amount of spatial colour is given, and the reliability of the decision can thus be illustrated for example by formulating a value I indicating the degree of the colour of the interference as follows:
The value I for the degree of the colour of the interference can be monitored and the value I for the degree of colour of the interference can be signalled to the radio network management system, from where the operator, for example, receives the information and can thus influence the maintenance and design of the radio network system. As already stated above, the changing of the combining method involves the changing of the covariance matrix model R_{rr }when a weight vector is to be formed for weighting the signals of the different antenna elements before they are combined. The covariance matrix R_{rr }estimates are shown in Table 2. When the MRC method is being used, the weight vector is thus formed as follows:
The term
preceding the channel estimate is a scalar coefficient the value of which may be assumed to be 1, or any other fixed value, and therefore it does not need to be calculated if the MRC method is used all the time. When the IRC method is used, the weight vector w_{k,m }is formed as follows:
w _{k,m} =c{circumflex over (R)} _{rr} ^{−1} h _{k,m} (26)
where c is a scalar coefficient that may obtain the value 1, {circumflex over (R)}_{rr }is a covariance matrix, h_{k,m }is a channel estimate, k is a user index and m is a path index. A separate weight vector w_{k,m }is formed for each delay to be examined. The length of the weight vector is L elements, L being the number of the antenna elements.
The weight coefficients can be formed by using an estimate of covariance matrix estimate {circumflex over (R)}_{rr }and a channel estimate by applying for example the LMS (Least Mean Square) and DMI (Direct Matrix Inversion) algorithms, which are described in greater detail in Winters, J., H., Signal Acquisition and Tracking with Adaptive Arrays in Wireless Systems, pp. 85–88, IEEE, 1993, to be included herein by reference. An aspect that slightly simplifies the solution is that the covariance matrix {circumflex over (R)}_{rr }is the same for all cell users and delays. The combining method is changed and scaled preferably framebyframe. A 3GPP (Third Generation Partnership Project) detector, together with a demodulator, carries out the soft decision of the received symbol, which may be a bit or a bit combination. The output variables of the soft decision detector are called soft symbols or bits. The soft symbols are decoded in a channel decoder. According to the 3GPP standard, convolution coding or turbo channel coding is used. When MRC combining is changed to IRC combining, the weighting of the soft bits should remain unchanged, otherwise the decoder cannot function properly.
The IRC method occupies more baseband processing capacity than the MRC method. When used extensively, baseband processing may reduce the ability of the base station to serve subscriber terminals and thereby impair cell capacity. The use of the IRC method may even impair the performance of the receiver due to errors in parameter estimation, because the number of parameters to be estimated is large and the more parameters are estimated, the higher is error probability. In the MRC method there are fewer parameters to be estimated than in the IRC method. Consequently, it is reasonable to use the IRC method only when there is a real need for it.
With reference to
Each Rake branch 212A–212B comprises an estimator 216 to which a baseband signal is supplied from each antenna element 200–202. The estimator 216 forms a channel estimate h and covariance matrix R_{uu}≈R_{rr }for the signal received by each antenna element 200–202. The spatial colour of the interference is measured at an interference detector 217, which controls the estimator 216 in accordance with the disclosed method to form complex weight coefficients w_{1}–w_{L }for the signals received by the different antenna elements 200–202. The estimator 216 functions as a controlling means, the weight coefficients w_{1}–w_{L }formed by it being used for multiplying signals arriving from the different antenna elements 200–202 in multipliers 220. From the interference detector 217, information about the spatial colour of the interference can be supplied further to monitoring means 219 in the radio system management system. In the areas encircled by a dotted line, the processing 218A of a pilot signal contained in the signal and the processing 218B of the data part contained in the signal take place. An antenna branch summer 222A located at the end of the Rake branch 212A–212B is used for combining the weighted pilot signal components to form a single pilot signal. Correspondingly, data signal components are combined in an antenna branch summer 222B to form a single data signal. The Rake receiver further comprises a Rake branch summer 224A, which is used for combining the pilot signals of Rake branches 212A–121B operating at different delays to form a summed pilot signal. The summed pilot signal can be further supplied to a signaltointerference ratio estimator 226, in which the signaltointerference ratio of the channel in question is estimated. The obtained signaltointerference ratio of the channel can be used for controlling closed loop power supply. The Rake branch summer 224B, in turn, is used for combining the data signals of the Rake branches 212A–212B operating at different delays to form a summed data signal.
If the source of spatially coloured interference is an illegal or malfunctioning transmitter, the disclosed Solution allows the interference to be reduced or removed. This can be achieved by means of different combining methods, by locating the source of the interference, by switching off the interfering transmitter, or by reducing the intensity of the interference of the transmitter.
The disclosed receiver is applicable both in a CDMA base station and subscriber terminal. The computation associated with the covariance matrix {circumflex over (R)}_{rr }can be carried out using an ASIC (Application Specific Integrated Circuit) to ensure that the computation does not load the DSP (Digital Signal Processing) resources, although the computational operations associated with the disclosed solution can be carried out using DSP software. For a precise covariance estimate to be produced, the sampling should preferably be carried out at the chip frequency of the spreading code. It is also possible to form the covariance matrix from samples taken at a lower frequency, the received sample sequence being decimated by N and only each N^{th }sample being used.
Although the invention is disclosed above with reference to an example based on the accompanying drawings, it is apparent that the invention is not restricted thereto but can be varied in many ways within the inventive idea disclosed in the accompanying claims.
Cited Patent  Filing date  Publication date  Applicant  Title 

US6128355  May 21, 1997  Oct 3, 2000  Telefonaktiebolget Lm Ericsson  Selective diversity combining 
US6144711  Aug 27, 1997  Nov 7, 2000  Cisco Systems, Inc.  Spatiotemporal processing for communication 
US6215814 *  May 15, 2000  Apr 10, 2001  Nokia Networks Oy  RAKE receiver 
US6529545 *  Mar 30, 2001  Mar 4, 2003  Nokia Networks Oy  Rake receiver 
EP1150441A1  Apr 25, 2000  Oct 31, 2001  Alcatel Alsthom Compagnie Generale D'electricite  Method of combining at least two received signals of a telecommunication system 
WO2000016494A1  Sep 14, 1999  Mar 23, 2000  Nokia Networks Oy  Rake receiver 
WO2000036765A1  Dec 15, 1998  Jun 22, 2000  Nokia Networks Oy  A receiver 
Reference  

1  "Signal Acquisition and Tracking with Adaptive Arrays in Wireless Systems", Jack H. Winters, IEEE, 1993, pp. 8588. 
Citing Patent  Filing date  Publication date  Applicant  Title 

US7627059  Apr 5, 2005  Dec 1, 2009  Samsung Electronics Co., Ltd.  Method of robust timing detection and carrier frequency offset estimation for OFDM systems 
US7715806 *  Apr 13, 2006  May 11, 2010  Broadcom Corporation  Method and system for diversity processing including using dedicated pilot method for closed loop 
US7751506  Dec 1, 2005  Jul 6, 2010  Samsung Electronics Co., Ltd.  Method for the soft bit metric calculation with linear MIMO detection for LDPC codes 
US7904039 *  Jan 31, 2005  Mar 8, 2011  UNIVERSITé LAVAL  Multiuser adaptive array receiver and method 
US8023907  Mar 25, 2010  Sep 20, 2011  Broadcom Corporation  Method and system for diversity processing including using dedicated pilot method for open loop 
US8320848  Mar 23, 2010  Nov 27, 2012  Broadcom Corporation  Method and system for diversity processing including using dedicated pilot method for open loop 
US8798654 *  Apr 22, 2009  Aug 5, 2014  Nokia Siemens Networks Oy  Selective interference rejection combining 
US9667372 *  Sep 26, 2013  May 30, 2017  Ntt Docomo, Inc.  Mobile communication terminal 
US20060183439 *  Apr 13, 2006  Aug 17, 2006  Vinko Erceg  Method and system for diversity processing including using dedicated pilot method for closed loop 
US20060222095 *  Apr 5, 2005  Oct 5, 2006  Samsung Electronics Co., Ltd.  Method of robust timing detection and carrier frequency offset estimation for OFDM systems 
US20070142089 *  Jan 31, 2005  Jun 21, 2007  Roy Sebastien Joseph A  Multiuser adaptive array receiver and method 
US20100172304 *  Mar 23, 2010  Jul 8, 2010  Broadcom Corporation  Method and system for diversity processing including using dedicated pilot method for open loop 
US20120115520 *  Apr 22, 2009  May 10, 2012  Roessel Sabine  Selective Interference Rejection Combining 
US20150358104 *  Sep 26, 2013  Dec 10, 2015  Ntt Docomo, Inc.  Mobile communication terminal 
U.S. Classification  375/347 
International Classification  H04B7/10, H04B7/08 
Cooperative Classification  H04B7/0857, H04B7/0871 
European Classification  H04B7/08H1, H04B7/08C4J3 
Date  Code  Event  Description 

Nov 14, 2003  AS  Assignment  Owner name: NOKIA CORPORATION, FINLAND Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNORS:ARTAMO, ATTE;ASTELY, DAVID;REEL/FRAME:014695/0252;SIGNING DATES FROM 20030903 TO 20030908 
Nov 10, 2010  FPAY  Fee payment  Year of fee payment: 4 
Nov 13, 2014  FPAY  Fee payment  Year of fee payment: 8 
Feb 17, 2015  AS  Assignment  Owner name: III HOLDINGS 3, LLC, DELAWARE Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:NOKIA CORPORATION;REEL/FRAME:034974/0644 Effective date: 20150115 